Skip to content

Latest commit

 

History

History
200 lines (130 loc) · 4.59 KB

notes.md

File metadata and controls

200 lines (130 loc) · 4.59 KB

Theo

Moderator for the session

TODO

  • Send presentation to Bette
  • Write script for what to say on each slide
  • Rehearse and time presentation. 3x

Maybe

  • Make overlays have margins on each side. Box style

Planning

Format

4 minutes lighting talk. 4 slides a 1 minute. Or up to 8 slides a 30 seconds

High pace, high energy. Simple and focused message No questions slot

Outline

  • Introduction. Talk title, Name
  • Roest. What they do
  • Soundsensing. What we do.
  • Coffee roasting. Importance
  • First crack. What is it, why track it
  • Acoustic detection
  • Demo. Video showing action
  • CTA. Roest, Soundsensing

Messaging

Roest is world leader in modern coffee roasters. Among the first to ship TinyML sound sensing

Soundsensing is world expert in Audio Machine Learning, including on-edge processing and TinyML

Together have developed and delivered a working solution for a practical problem, using TinyML.

"this is example of something one can today with modern ML on Audio"

Soundsensing offering

Experts on Audio Machine Learning specializing in continious monitoring and on-edge Machine Learning

Hardware + Software

Privacy by design

On-edge Hybrid edge/cloud

Supporting

  • Sound Event Detection
  • Audio Classification
  • Anomaly Detection

Noise Monitoring for industrial, office and smart-city Condition Monitoring of Technical machinery

Audio Machine Learning models

Roasting coffee

Critical to terminate the roasting process at the same time One can do this

How we did it

Technology stack

Microphone. VM3000 digital MEMS by Vesper Microcontroller. STM32L476 by STMicroelectronics Inference engine. X-CUBE-AI by STMicroelectronics Machine Learning. Keras+Tensorflow with Soundsensing SDK Firmware. Custom, using Soundsensing SDK Hardware. Custom board by Soundsensing

Call-to-Actions

If you want the worlds most advanced coffe-roaster, Contact Roest!

Have an sensing/monitoring problem that can be approached with sound? In process-, manufacturing or other industries. Contact Soundsensing!

contact@soundsensing.no

Maybe

Can be used to monitor quality. Can be used for automation. Saving time.

More consistent results. Variation in beans composition, moisture

Sampling rod. Glass window. Roasting profile. Timing.

https://www.roestcoffee.com/first-crack

Tilbakemeldinger fra Roest

Why acoustic first crack?

first crack essensielt en person 1 minutt etter first crack. Reduser power

100k maskiner out there. Operators listen manually for first crack. Requires being present and alert all the time. By, so that operator can do other things. Prepare coffee Evaluate coffee Operate more machines.

Sample roasters are used in the coffee production value chain to ensure quality. When coffee beans are sold/purcaseds

Sample 100 gram coffee At the farm At the outgoing distributors At incoming distributors Repacking

When you or your local coffee show buy quality coffee beans from reputable brands Beans from the same batch has been sampled, roasted, cupped and scored multiple times Sample roaster

https://www.roestcoffee.com/first-crack

variasjon

Roest som selskap. Ekstemt verdifult Tester 10 maskiner Internal quality assurance

ST. Mikro

Press-release info

Roest is a leading manufacturer of coffee-roasting machines. Their award-winning sample roaster is used by high-end coffee makers all over the world, and is the first commercially available coffee roaster that uses acoustic sensing.

Soundsensing is a leader in acoustic monitoring using Machine Learning. Their IoT sensors are used to monitor noise levels in workplaces and cities, and to monitor the condition of machines and processes in manufacturing and real-estate.

The model runs on a custom developed board using an STM32 microcontroller and the X-CUBE-AI machine learning toolkit. Roest coffee machines have been shipping with this TinyML-powered feature since August 2020.

Soundbites

Who here likes Coffee? Who here likes Machine Learning? Who here likes Microcontrollers? Then you have come to the right talk!

Did you know that roasting coffee beans make them crack like popcorn?

During the roasting process the coffebeans heat up and the water inside turns to steam. This can be released abruptly through the hard shell, resulting in an audible "crack".

Experienced coffee roasters often listen for these cracks. But this requires constant virgilance, making it hard to do other tasks at the same time.

But how do coffee-makers select their coffee? Testing the coffe. Roasting chart.

Development process. From the start and all the way to a validated solution, system design, data collection, model development, and testing.